Split out hash functions into new file and turn into inline static functions

This includes various tweaks related to building SipHash and other
cleanup.

1 files changed, 2 insertions, 572 deletions

diff --git a/hv.h b/hv.h
index 3ee2399ea4..dddeb027e9 100644
--- a/hv.h
+++ b/hv.h
@@ -103,578 +103,6 @@ struct xpvhv {
     STRLEN      xhv_max;        /* subscript of last element of xhv_array */
 };
 
-/* hash a key */
-/* The use of a temporary pointer and the casting games
- * is needed to serve the dual purposes of
- * (a) the hashed data being interpreted as "unsigned char" (new since 5.8,
- *     a "char" can be either signed or unsigned, depending on the compiler)
- * (b) catering for old code that uses a "char"
- *
- * The "hash seed" feature was added in Perl 5.8.1 to perturb the results
- * to avoid "algorithmic complexity attacks".
- *
- * If USE_HASH_SEED is defined, hash randomisation is done by default
- * If USE_HASH_SEED_EXPLICIT is defined, hash randomisation is done
- * only if the environment variable PERL_HASH_SEED is set.
- * (see also perl.c:perl_parse() and S_init_tls_and_interp() and util.c:get_hash_seed())
- */
-#ifndef PERL_HASH_SEED
-#   if defined(USE_HASH_SEED) || defined(USE_HASH_SEED_EXPLICIT)
-#       define PERL_HASH_SEED PL_hash_seed
-#   else
-#       define PERL_HASH_SEED "PeRlHaShhAcKpErl"
-#   endif
-#endif
-
-#define PERL_HASH_SEED_U32   *((U32*)PERL_HASH_SEED)
-#define PERL_HASH_SEED_U64_1 (((U64*)PERL_HASH_SEED)[0])
-#define PERL_HASH_SEED_U64_2 (((U64*)PERL_HASH_SEED)[1])
-#define PERL_HASH_SEED_U16_x(idx) (((U16*)PERL_HASH_SEED)[idx])
-
-/* legacy - only mod_perl should be doing this.  */
-#ifdef PERL_HASH_INTERNAL_ACCESS
-#define PERL_HASH_INTERNAL(hash,str,len) PERL_HASH(hash,str,len)
-#endif
-
-/* Uncomment one of the following lines to use an alternative hash algorithm.
-#define PERL_HASH_FUNC_SDBM
-#define PERL_HASH_FUNC_DJB2
-#define PERL_HASH_FUNC_SUPERFAST
-#define PERL_HASH_FUNC_MURMUR3
-#define PERL_HASH_FUNC_SIPHASH
-#define PERL_HASH_FUNC_ONE_AT_A_TIME
-#define PERL_HASH_FUNC_ONE_AT_A_TIME_OLD
-#define PERL_HASH_FUNC_BUZZHASH16
-*/
-
-#if !( 0 \
-        || defined(PERL_HASH_FUNC_SDBM) \
-        || defined(PERL_HASH_FUNC_DJB2) \
-        || defined(PERL_HASH_FUNC_SUPERFAST) \
-        || defined(PERL_HASH_FUNC_MURMUR3) \
-        || defined(PERL_HASH_FUNC_ONE_AT_A_TIME) \
-        || defined(PERL_HASH_FUNC_ONE_AT_A_TIME_OLD) \
-        || defined(PERL_HASH_FUNC_BUZZHASH16) \
-    )
-#ifdef U64
-#define PERL_HASH_FUNC_SIPHASH
-#else
-#define PERL_HASH_FUNC_ONE_AT_A_TIME
-#endif
-#endif
-
-#if defined(PERL_HASH_FUNC_BUZZHASH16)
-/* "BUZZHASH16"
- *
- * I whacked this together while just playing around.
- *
- * The idea is that instead of hashing the actual string input we use the
- * bytes of the string as an index into a table of randomly generated
- * 16 bit values.
- *
- * A left rotate is used to "mix" in previous bits as we go, and I borrowed
- * the avalanche function from one-at-a-time for the final step. A lookup
- * into the table based on the lower 8 bits of the length combined with
- * the length itself is used as an itializer.
- *
- * The resulting hash value has no actual bits fed in from the string so
- * I would guess it is pretty secure, although I am not a cryptographer
- * and have no idea for sure. Nor has it been rigorously tested. On the
- * other hand it is reasonably fast, and seems to produce reasonable
- * distributions.
- *
- * Yves Orton
- */
-
-
-#define PERL_HASH_FUNC "BUZZHASH16"
-#define PERL_HASH_SEED_BYTES 512 /* 2 bytes per octet value, 2 * 256 */
-/* Find best way to ROTL32 */
-#if defined(_MSC_VER)
-  #include <stdlib.h>  /* Microsoft put _rotl declaration in here */
-  #define BUZZHASH_ROTL32(x,r)  _rotl(x,r)
-#else
-  /* gcc recognises this code and generates a rotate instruction for CPUs with one */
-  #define BUZZHASH_ROTL32(x,r)  (((U32)x << r) | ((U32)x >> (32 - r)))
-#endif
-
-#define PERL_HASH(hash,str,len) \
-     STMT_START        { \
-        const char * const s_PeRlHaSh_tmp = (str); \
-        const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
-        const unsigned char *end_PeRlHaSh = (const unsigned char *)s_PeRlHaSh + len; \
-        U32 hash_PeRlHaSh = (PERL_HASH_SEED_U16_x(len & 0xff) << 16) + len; \
-        while (s_PeRlHaSh < end_PeRlHaSh) { \
-            hash_PeRlHaSh ^= PERL_HASH_SEED_U16_x((U8)*s_PeRlHaSh++); \
-            hash_PeRlHaSh += BUZZHASH_ROTL32(hash_PeRlHaSh,11); \
-        } \
-        hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
-        hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
-        (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
-    } STMT_END
-
-#elif defined(PERL_HASH_FUNC_SIPHASH)
-#define PERL_HASH_FUNC "SIPHASH"
-#define PERL_HASH_SEED_BYTES 16
-
-/* This is SipHash by Jean-Philippe Aumasson and Daniel J. Bernstein.
- * The authors claim it is relatively secure compared to the alternatives
- * and that performance wise it is a suitable hash for languages like Perl.
- * See:
- *
- * https://www.131002.net/siphash/
- *
- * This implementation seems to perform slightly slower than one-at-a-time for
- * short keys, but degrades slower for longer keys. Murmur Hash outperforms it
- * regardless of keys size.
- *
- * It is 64 bit only.
- */
-
-#define PERL_HASH_NEEDS_TWO_SEEDS
-
-#ifndef U64
-#define U64 uint64_t
-#endif
-
-#define ROTL(x,b) (U64)( ((x) << (b)) | ( (x) >> (64 - (b))) )
-
-#define U32TO8_LE(p, v)         \
-    (p)[0] = (U8)((v)      ); (p)[1] = (U8)((v) >>  8); \
-    (p)[2] = (U8)((v) >> 16); (p)[3] = (U8)((v) >> 24);
-
-#define U64TO8_LE(p, v)         \
-  U32TO8_LE((p),     (U32)((v)      ));   \
-  U32TO8_LE((p) + 4, (U32)((v) >> 32));
-
-#define U8TO64_LE(p) \
-  (((U64)((p)[0])      ) | \
-   ((U64)((p)[1]) <<  8) | \
-   ((U64)((p)[2]) << 16) | \
-   ((U64)((p)[3]) << 24) | \
-   ((U64)((p)[4]) << 32) | \
-   ((U64)((p)[5]) << 40) | \
-   ((U64)((p)[6]) << 48) | \
-   ((U64)((p)[7]) << 56))
-
-#define SIPROUND            \
-  do {              \
-    v0_PeRlHaSh += v1_PeRlHaSh; v1_PeRlHaSh=ROTL(v1_PeRlHaSh,13); v1_PeRlHaSh ^= v0_PeRlHaSh; v0_PeRlHaSh=ROTL(v0_PeRlHaSh,32); \
-    v2_PeRlHaSh += v3_PeRlHaSh; v3_PeRlHaSh=ROTL(v3_PeRlHaSh,16); v3_PeRlHaSh ^= v2_PeRlHaSh;     \
-    v0_PeRlHaSh += v3_PeRlHaSh; v3_PeRlHaSh=ROTL(v3_PeRlHaSh,21); v3_PeRlHaSh ^= v0_PeRlHaSh;     \
-    v2_PeRlHaSh += v1_PeRlHaSh; v1_PeRlHaSh=ROTL(v1_PeRlHaSh,17); v1_PeRlHaSh ^= v2_PeRlHaSh; v2_PeRlHaSh=ROTL(v2_PeRlHaSh,32); \
-  } while(0)
-
-/* SipHash-2-4 */
-#define PERL_HASH(hash,str,len) STMT_START { \
-  const char * const strtmp_PeRlHaSh = (str); \
-  const unsigned char *in_PeRlHaSh = (const unsigned char *)strtmp_PeRlHaSh; \
-  const U32 inlen_PeRlHaSh = (len); \
-  /* "somepseudorandomlygeneratedbytes" */ \
-  U64 v0_PeRlHaSh = 0x736f6d6570736575ULL; \
-  U64 v1_PeRlHaSh = 0x646f72616e646f6dULL; \
-  U64 v2_PeRlHaSh = 0x6c7967656e657261ULL; \
-  U64 v3_PeRlHaSh = 0x7465646279746573ULL; \
-\
-  U64 b_PeRlHaSh;                           \
-  U64 k0_PeRlHaSh = PERL_HASH_SEED_U64_1;   \
-  U64 k1_PeRlHaSh = PERL_HASH_SEED_U64_2;   \
-  U64 m_PeRlHaSh;                           \
-  const int left_PeRlHaSh = inlen_PeRlHaSh & 7; \
-  const U8 *end_PeRlHaSh = in_PeRlHaSh + inlen_PeRlHaSh - left_PeRlHaSh; \
-\
-  b_PeRlHaSh = ( ( U64 )(len) ) << 56; \
-  v3_PeRlHaSh ^= k1_PeRlHaSh; \
-  v2_PeRlHaSh ^= k0_PeRlHaSh; \
-  v1_PeRlHaSh ^= k1_PeRlHaSh; \
-  v0_PeRlHaSh ^= k0_PeRlHaSh; \
-\
-  for ( ; in_PeRlHaSh != end_PeRlHaSh; in_PeRlHaSh += 8 ) \
-  { \
-    m_PeRlHaSh = U8TO64_LE( in_PeRlHaSh ); \
-    v3_PeRlHaSh ^= m_PeRlHaSh; \
-    SIPROUND; \
-    SIPROUND; \
-    v0_PeRlHaSh ^= m_PeRlHaSh; \
-  } \
-\
-  switch( left_PeRlHaSh ) \
-  { \
-  case 7: b_PeRlHaSh |= ( ( U64 )in_PeRlHaSh[ 6] )  << 48; \
-  case 6: b_PeRlHaSh |= ( ( U64 )in_PeRlHaSh[ 5] )  << 40; \
-  case 5: b_PeRlHaSh |= ( ( U64 )in_PeRlHaSh[ 4] )  << 32; \
-  case 4: b_PeRlHaSh |= ( ( U64 )in_PeRlHaSh[ 3] )  << 24; \
-  case 3: b_PeRlHaSh |= ( ( U64 )in_PeRlHaSh[ 2] )  << 16; \
-  case 2: b_PeRlHaSh |= ( ( U64 )in_PeRlHaSh[ 1] )  <<  8; \
-  case 1: b_PeRlHaSh |= ( ( U64 )in_PeRlHaSh[ 0] ); break; \
-  case 0: break; \
-  } \
-\
-  v3_PeRlHaSh ^= b_PeRlHaSh; \
-  SIPROUND; \
-  SIPROUND; \
-  v0_PeRlHaSh ^= b_PeRlHaSh; \
-\
-  v2_PeRlHaSh ^= 0xff; \
-  SIPROUND; \
-  SIPROUND; \
-  SIPROUND; \
-  SIPROUND; \
-  b_PeRlHaSh = v0_PeRlHaSh ^ v1_PeRlHaSh ^ v2_PeRlHaSh  ^ v3_PeRlHaSh; \
-  (hash)= (U32)(b_PeRlHaSh & U32_MAX); \
-} STMT_END
-
-#elif defined(PERL_HASH_FUNC_SUPERFAST)
-#define PERL_HASH_FUNC "SUPERFAST"
-#define PERL_HASH_SEED_BYTES 4
-/* FYI: This is the "Super-Fast" algorithm mentioned by Bob Jenkins in
- * (http://burtleburtle.net/bob/hash/doobs.html)
- * It is by Paul Hsieh (c) 2004 and is analysed here
- * http://www.azillionmonkeys.com/qed/hash.html
- * license terms are here:
- * http://www.azillionmonkeys.com/qed/weblicense.html
- */
-#undef get16bits
-#if (defined(__GNUC__) && defined(__i386__)) || defined(__WATCOMC__) \
-  || defined(_MSC_VER) || defined (__BORLANDC__) || defined (__TURBOC__)
-#define get16bits(d) (*((const U16 *) (d)))
-#endif
-
-#if !defined (get16bits)
-#define get16bits(d) ((((const U8 *)(d))[1] << UINT32_C(8))\
-                      +((const U8 *)(d))[0])
-#endif
-#define PERL_HASH(hash,str,len) \
-      STMT_START        { \
-        const char * const strtmp_PeRlHaSh = (str); \
-        const unsigned char *str_PeRlHaSh = (const unsigned char *)strtmp_PeRlHaSh; \
-        U32 len_PeRlHaSh = (len); \
-        U32 hash_PeRlHaSh = PERL_HASH_SEED_U32 ^ len; \
-        U32 tmp_PeRlHaSh; \
-        int rem_PeRlHaSh= len_PeRlHaSh & 3; \
-        len_PeRlHaSh >>= 2; \
-                            \
-        for (;len_PeRlHaSh > 0; len_PeRlHaSh--) { \
-            hash_PeRlHaSh  += get16bits (str_PeRlHaSh); \
-            tmp_PeRlHaSh    = (get16bits (str_PeRlHaSh+2) << 11) ^ hash_PeRlHaSh; \
-            hash_PeRlHaSh   = (hash_PeRlHaSh << 16) ^ tmp_PeRlHaSh; \
-            str_PeRlHaSh   += 2 * sizeof (U16); \
-            hash_PeRlHaSh  += hash_PeRlHaSh >> 11; \
-        } \
-        \
-        /* Handle end cases */ \
-        switch (rem_PeRlHaSh) { \
-            case 3: hash_PeRlHaSh += get16bits (str_PeRlHaSh); \
-                    hash_PeRlHaSh ^= hash_PeRlHaSh << 16; \
-                    hash_PeRlHaSh ^= str_PeRlHaSh[sizeof (U16)] << 18; \
-                    hash_PeRlHaSh += hash_PeRlHaSh >> 11; \
-                    break; \
-            case 2: hash_PeRlHaSh += get16bits (str_PeRlHaSh); \
-                    hash_PeRlHaSh ^= hash_PeRlHaSh << 11; \
-                    hash_PeRlHaSh += hash_PeRlHaSh >> 17; \
-                    break; \
-            case 1: hash_PeRlHaSh += *str_PeRlHaSh; \
-                    hash_PeRlHaSh ^= hash_PeRlHaSh << 10; \
-                    hash_PeRlHaSh += hash_PeRlHaSh >> 1; \
-        } \
-        \
-        /* Force "avalanching" of final 127 bits */ \
-        hash_PeRlHaSh ^= hash_PeRlHaSh << 3; \
-        hash_PeRlHaSh += hash_PeRlHaSh >> 5; \
-        hash_PeRlHaSh ^= hash_PeRlHaSh << 4; \
-        hash_PeRlHaSh += hash_PeRlHaSh >> 17; \
-        hash_PeRlHaSh ^= hash_PeRlHaSh << 25; \
-        (hash) = (hash_PeRlHaSh + (hash_PeRlHaSh >> 6)); \
-    } STMT_END
-
-#elif defined(PERL_HASH_FUNC_MURMUR3)
-#define PERL_HASH_FUNC "MURMUR3"
-#define PERL_HASH_SEED_BYTES 4
-
-/*-----------------------------------------------------------------------------
- * MurmurHash3 was written by Austin Appleby, and is placed in the public
- * domain.
- *
- * This implementation was originally written by Shane Day, and is also public domain,
- * and was modified to function as a macro similar to other perl hash functions by
- * Yves Orton.
- *
- * This is a portable ANSI C implementation of MurmurHash3_x86_32 (Murmur3A)
- * with support for progressive processing.
- *
- * If you want to understand the MurmurHash algorithm you would be much better
- * off reading the original source. Just point your browser at:
- * http://code.google.com/p/smhasher/source/browse/trunk/MurmurHash3.cpp
- *
- * How does it work?
- *
- * We can only process entire 32 bit chunks of input, except for the very end
- * that may be shorter.
- *
- * To handle endianess I simply use a macro that reads a U32 and define
- * that macro to be a direct read on little endian machines, a read and swap
- * on big endian machines, or a byte-by-byte read if the endianess is unknown.
- */
-
-
-/*-----------------------------------------------------------------------------
- * Endianess, misalignment capabilities and util macros
- *
- * The following 3 macros are defined in this section. The other macros defined
- * are only needed to help derive these 3.
- *
- * MURMUR_READ_UINT32(x)   Read a little endian unsigned 32-bit int
- * MURMUR_UNALIGNED_SAFE   Defined if READ_UINT32 works on non-word boundaries
- * MURMUR_ROTL32(x,r)      Rotate x left by r bits
- */
-
-/* Now find best way we can to READ_UINT32 */
-#if (BYTEORDER == 0x1234 || BYTEORDER == 0x12345678) && U32SIZE == 4
-  /* CPU endian matches murmurhash algorithm, so read 32-bit word directly */
-  #define MURMUR_READ_UINT32(ptr)   (*((U32*)(ptr)))
-#elif BYTEORDER == 0x4321 || BYTEORDER == 0x87654321
-  /* TODO: Add additional cases below where a compiler provided bswap32 is available */
-  #if defined(__GNUC__) && (__GNUC__>4 || (__GNUC__==4 && __GNUC_MINOR__>=3))
-    #define MURMUR_READ_UINT32(ptr)   (__builtin_bswap32(*((U32*)(ptr))))
-  #else
-    /* Without a known fast bswap32 we're just as well off doing this */
-    #define MURMUR_READ_UINT32(ptr)   (ptr[0]|ptr[1]<<8|ptr[2]<<16|ptr[3]<<24)
-    #define MURMUR_UNALIGNED_SAFE
-  #endif
-#else
-  /* Unknown endianess so last resort is to read individual bytes */
-  #define MURMUR_READ_UINT32(ptr)   (ptr[0]|ptr[1]<<8|ptr[2]<<16|ptr[3]<<24)
-
-  /* Since we're not doing word-reads we can skip the messing about with realignment */
-  #define MURMUR_UNALIGNED_SAFE
-#endif
-
-/* Find best way to ROTL32 */
-#if defined(_MSC_VER)
-  #include <stdlib.h>  /* Microsoft put _rotl declaration in here */
-  #define MURMUR_ROTL32(x,r)  _rotl(x,r)
-#else
-  /* gcc recognises this code and generates a rotate instruction for CPUs with one */
-  #define MURMUR_ROTL32(x,r)  (((U32)x << r) | ((U32)x >> (32 - r)))
-#endif
-
-
-/*-----------------------------------------------------------------------------
- * Core murmurhash algorithm macros */
-
-#define MURMUR_C1  (0xcc9e2d51)
-#define MURMUR_C2  (0x1b873593)
-#define MURMUR_C3  (0xe6546b64)
-#define MURMUR_C4  (0x85ebca6b)
-#define MURMUR_C5  (0xc2b2ae35)
-
-/* This is the main processing body of the algorithm. It operates
- * on each full 32-bits of input. */
-#define MURMUR_DOBLOCK(h1, k1) STMT_START { \
-    k1 *= MURMUR_C1; \
-    k1 = MURMUR_ROTL32(k1,15); \
-    k1 *= MURMUR_C2; \
-    \
-    h1 ^= k1; \
-    h1 = MURMUR_ROTL32(h1,13); \
-    h1 = h1 * 5 + MURMUR_C3; \
-} STMT_END
-
-
-/* Append unaligned bytes to carry, forcing hash churn if we have 4 bytes */
-/* cnt=bytes to process, h1=name of h1 var, c=carry, n=bytes in c, ptr/len=payload */
-#define MURMUR_DOBYTES(cnt, h1, c, n, ptr, len) STMT_START { \
-    int MURMUR_DOBYTES_i = cnt; \
-    while(MURMUR_DOBYTES_i--) { \
-        c = c>>8 | *ptr++<<24; \
-        n++; len--; \
-        if(n==4) { \
-            MURMUR_DOBLOCK(h1, c); \
-            n = 0; \
-        } \
-    } \
-} STMT_END
-
-/* process the last 1..3 bytes and finalize */
-#define MURMUR_FINALIZE(hash, PeRlHaSh_len, PeRlHaSh_k1, PeRlHaSh_h1, PeRlHaSh_carry, PeRlHaSh_bytes_in_carry, PeRlHaSh_ptr, PeRlHaSh_total_length) STMT_START { \
-    /* Advance over whole 32-bit chunks, possibly leaving 1..3 bytes */\
-    PeRlHaSh_len -= PeRlHaSh_len/4*4;                           \
-                                                                \
-    /* Append any remaining bytes into carry */                 \
-    MURMUR_DOBYTES(PeRlHaSh_len, PeRlHaSh_h1, PeRlHaSh_carry, PeRlHaSh_bytes_in_carry, PeRlHaSh_ptr, PeRlHaSh_len); \
-                                                                \
-    if (PeRlHaSh_bytes_in_carry) {                                           \
-        PeRlHaSh_k1 = PeRlHaSh_carry >> ( 4 - PeRlHaSh_bytes_in_carry ) * 8; \
-        PeRlHaSh_k1 *= MURMUR_C1;                               \
-        PeRlHaSh_k1 = MURMUR_ROTL32(PeRlHaSh_k1,15);                   \
-        PeRlHaSh_k1 *= MURMUR_C2;                               \
-        PeRlHaSh_h1 ^= PeRlHaSh_k1;                             \
-    }                                                           \
-    PeRlHaSh_h1 ^= PeRlHaSh_total_length;                       \
-                                                                \
-    /* fmix */                                                  \
-    PeRlHaSh_h1 ^= PeRlHaSh_h1 >> 16;                           \
-    PeRlHaSh_h1 *= MURMUR_C4;                                   \
-    PeRlHaSh_h1 ^= PeRlHaSh_h1 >> 13;                           \
-    PeRlHaSh_h1 *= MURMUR_C5;                                   \
-    PeRlHaSh_h1 ^= PeRlHaSh_h1 >> 16;                           \
-    (hash)= PeRlHaSh_h1;                                        \
-} STMT_END
-
-/* now we create the hash function */
-
-#if defined(UNALIGNED_SAFE)
-#define PERL_HASH(hash,str,len) STMT_START { \
-        const char * const s_PeRlHaSh_tmp = (str); \
-        const unsigned char *PeRlHaSh_ptr = (const unsigned char *)s_PeRlHaSh_tmp; \
-        I32 PeRlHaSh_len = len;    \
-                                            \
-        U32 PeRlHaSh_h1 = PERL_HASH_SEED_U32;   \
-        U32 PeRlHaSh_k1;                    \
-        U32 PeRlHaSh_carry = 0;             \
-                                            \
-        const unsigned char *PeRlHaSh_end;  \
-                                            \
-        int PeRlHaSh_bytes_in_carry = 0; /* bytes in carry */ \
-        I32 PeRlHaSh_total_length= PeRlHaSh_len; \
-                                            \
-        /* This CPU handles unaligned word access */            \
-        /* Process 32-bit chunks */                             \
-        PeRlHaSh_end = PeRlHaSh_ptr + PeRlHaSh_len/4*4;         \
-        for( ; PeRlHaSh_ptr < PeRlHaSh_end ; PeRlHaSh_ptr+=4) { \
-            PeRlHaSh_k1 = MURMUR_READ_UINT32(PeRlHaSh_ptr);        \
-            MURMUR_DOBLOCK(PeRlHaSh_h1, PeRlHaSh_k1);                  \
-        }                                                       \
-        \
-        MURMUR_FINALIZE(hash, PeRlHaSh_len, PeRlHaSh_k1, PeRlHaSh_h1, PeRlHaSh_carry, PeRlHaSh_bytes_in_carry, PeRlHaSh_ptr, PeRlHaSh_total_length);\
-    } STMT_END
-#else
-#define PERL_HASH(hash,str,len) STMT_START { \
-        const char * const s_PeRlHaSh_tmp = (str); \
-        const unsigned char *PeRlHaSh_ptr = (const unsigned char *)s_PeRlHaSh_tmp; \
-        I32 PeRlHaSh_len = len;    \
-                                            \
-        U32 PeRlHaSh_h1 = PERL_HASH_SEED_U32;   \
-        U32 PeRlHaSh_k1;                    \
-        U32 PeRlHaSh_carry = 0;             \
-                                            \
-        const unsigned char *PeRlHaSh_end;  \
-                                            \
-        int PeRlHaSh_bytes_in_carry = 0; /* bytes in carry */ \
-        I32 PeRlHaSh_total_length= PeRlHaSh_len; \
-                                            \
-        /* This CPU does not handle unaligned word access */    \
-                                                                \
-        /* Consume enough so that the next data byte is word aligned */ \
-        int PeRlHaSh_i = -(long)PeRlHaSh_ptr & 3;                       \
-        if(PeRlHaSh_i && PeRlHaSh_i <= PeRlHaSh_len) {                  \
-          MURMUR_DOBYTES(PeRlHaSh_i, PeRlHaSh_h1, PeRlHaSh_carry, PeRlHaSh_bytes_in_carry, PeRlHaSh_ptr, PeRlHaSh_len);\
-        }                                                               \
-        \
-        /* We're now aligned. Process in aligned blocks. Specialise for each possible carry count */ \
-        PeRlHaSh_end = PeRlHaSh_ptr + PeRlHaSh_len/4*4;                 \
-        switch(PeRlHaSh_bytes_in_carry) { /* how many bytes in carry */                  \
-            case 0: /* c=[----]  w=[3210]  b=[3210]=w            c'=[----] */ \
-            for( ; PeRlHaSh_ptr < PeRlHaSh_end ; PeRlHaSh_ptr+=4) { \
-                PeRlHaSh_k1 = MURMUR_READ_UINT32(PeRlHaSh_ptr);        \
-                MURMUR_DOBLOCK(PeRlHaSh_h1, PeRlHaSh_k1);                  \
-            }                                                       \
-            break;                                                  \
-            case 1: /* c=[0---]  w=[4321]  b=[3210]=c>>24|w<<8   c'=[4---] */   \
-            for( ; PeRlHaSh_ptr < PeRlHaSh_end ; PeRlHaSh_ptr+=4) { \
-                PeRlHaSh_k1 = PeRlHaSh_carry>>24;                   \
-                PeRlHaSh_carry = MURMUR_READ_UINT32(PeRlHaSh_ptr);             \
-                PeRlHaSh_k1 |= PeRlHaSh_carry<<8;                       \
-                MURMUR_DOBLOCK(PeRlHaSh_h1, PeRlHaSh_k1);                  \
-            }                                                       \
-            break;                                                  \
-            case 2: /* c=[10--]  w=[5432]  b=[3210]=c>>16|w<<16  c'=[54--] */   \
-            for( ; PeRlHaSh_ptr < PeRlHaSh_end ; PeRlHaSh_ptr+=4) { \
-                PeRlHaSh_k1 = PeRlHaSh_carry>>16;                   \
-                PeRlHaSh_carry = MURMUR_READ_UINT32(PeRlHaSh_ptr);             \
-                PeRlHaSh_k1 |= PeRlHaSh_carry<<16;                      \
-                MURMUR_DOBLOCK(PeRlHaSh_h1, PeRlHaSh_k1);                  \
-            }                                                       \
-            break;                                                  \
-            case 3: /* c=[210-]  w=[6543]  b=[3210]=c>>8|w<<24   c'=[654-] */   \
-            for( ; PeRlHaSh_ptr < PeRlHaSh_end ; PeRlHaSh_ptr+=4) { \
-                PeRlHaSh_k1 = PeRlHaSh_carry>>8;                    \
-                PeRlHaSh_carry = MURMUR_READ_UINT32(PeRlHaSh_ptr);             \
-                PeRlHaSh_k1 |= PeRlHaSh_carry<<24;                      \
-                MURMUR_DOBLOCK(PeRlHaSh_h1, PeRlHaSh_k1);                  \
-            }                                                       \
-        }                                                           \
-                                                                    \
-        MURMUR_FINALIZE(hash, PeRlHaSh_len, PeRlHaSh_k1, PeRlHaSh_h1, PeRlHaSh_carry, PeRlHaSh_bytes_in_carry, PeRlHaSh_ptr, PeRlHaSh_total_length);\
-    } STMT_END
-#endif
-
-#elif defined(PERL_HASH_FUNC_DJB2)
-#define PERL_HASH_FUNC "DJB2"
-#define PERL_HASH_SEED_BYTES 4
-#define PERL_HASH(hash,str,len) \
-     STMT_START        { \
-        const char * const s_PeRlHaSh_tmp = (str); \
-        const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
-        I32 i_PeRlHaSh = len; \
-        U32 hash_PeRlHaSh = PERL_HASH_SEED_U32 ^ len; \
-        while (i_PeRlHaSh--) { \
-            hash_PeRlHaSh = ((hash_PeRlHaSh << 5) + hash_PeRlHaSh) + *s_PeRlHaSh++; \
-        } \
-        (hash) = hash_PeRlHaSh;\
-    } STMT_END
-
-#elif defined(PERL_HASH_FUNC_SDBM)
-#define PERL_HASH_FUNC "SDBM"
-#define PERL_HASH_SEED_BYTES 4
-#define PERL_HASH(hash,str,len) \
-     STMT_START        { \
-        const char * const s_PeRlHaSh_tmp = (str); \
-        const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
-        I32 i_PeRlHaSh = len; \
-        U32 hash_PeRlHaSh = PERL_HASH_SEED_U32 ^ len; \
-        while (i_PeRlHaSh--) { \
-            hash_PeRlHaSh = (hash_PeRlHaSh << 6) + (hash_PeRlHaSh << 16) - hash_PeRlHaSh + *s_PeRlHaSh++; \
-        } \
-        (hash) = hash_PeRlHaSh;\
-    } STMT_END
-
-#elif defined(PERL_HASH_FUNC_ONE_AT_A_TIME) || defined(PERL_HASH_FUNC_ONE_AT_A_TIME_OLD)
-
-#define PERL_HASH_SEED_BYTES 4
-
-#ifdef PERL_HASH_FUNC_ONE_AT_A_TIME
-/* new version, add the length to the seed so that adding characters changes the "seed" being used. */
-#define PERL_HASH_FUNC "ONE_AT_A_TIME"
-#define MIX_SEED_AND_LEN(seed,len) (seed + len)
-#else
-/* old version, just use the seed. - not recommended */
-#define PERL_HASH_FUNC "ONE_AT_A_TIME_OLD"
-#define MIX_SEED_AND_LEN(seed,len) (seed)
-#endif
-
-/* FYI: This is the "One-at-a-Time" algorithm by Bob Jenkins
- * from requirements by Colin Plumb.
- * (http://burtleburtle.net/bob/hash/doobs.html) */
-#define PERL_HASH(hash,str,len) \
-     STMT_START	{ \
-        const char * const s_PeRlHaSh_tmp = (str); \
-        const unsigned char *s_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp; \
-        const unsigned char *end_PeRlHaSh = (const unsigned char *)s_PeRlHaSh_tmp + (len); \
-        U32 hash_PeRlHaSh = MIX_SEED_AND_LEN(PERL_HASH_SEED_U32, len); \
-	while (s_PeRlHaSh < end_PeRlHaSh) { \
-            hash_PeRlHaSh += (U8)*s_PeRlHaSh++; \
-	    hash_PeRlHaSh += (hash_PeRlHaSh << 10); \
-	    hash_PeRlHaSh ^= (hash_PeRlHaSh >> 6); \
-	} \
-	hash_PeRlHaSh += (hash_PeRlHaSh << 3); \
-	hash_PeRlHaSh ^= (hash_PeRlHaSh >> 11); \
-	(hash) = (hash_PeRlHaSh + (hash_PeRlHaSh << 15)); \
-    } STMT_END
-#endif
-#ifndef PERL_HASH
-#error "No hash function defined!"
-#endif
 /*
 =head1 Hash Manipulation Functions
 
@@ -1131,6 +559,8 @@ Creates a new HV.  The reference count is set to 1.
 
 #define newHV()	MUTABLE_HV(newSV_type(SVt_PVHV))
 
+#include "hv_func.h"
+
 /*
  * Local variables:
  * c-indentation-style: bsd